PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 22, 20260 citationsOpen Access

AI for Direct Reporting of Ambulatory Electrocardiography

Artificial intelligence for direct-to-physician reporting of ambulatory electrocardiography

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Ambulatory ECG technology generates vast amounts of data currently requiring human technician interpretation, prompting the evaluation of an AI algorithm for direct-to-physician reporting.

Does an ensemble AI model (DeepRhythmAI) improve the identification of critical arrhythmias in ambulatory ECG recordings compared to certified ECG technicians?

Population

14,606 individual ambulatory ECG recordings

Comparison

DeepRhythmAI model vs 167 certified ECG technicians

Design

Comparative diagnostic accuracy study

Key result

DeepRhythmAI detected critical arrhythmias with 98.6% sensitivity versus 80.3% for technicians, reducing false-negatives by over 10-fold but increasing false-positives.

Authors

LJL S JohnsonPZP ZadrozniakGJG Jasina

Discussion

Loading...

Member takes

Overview

May support AI triage of ambulatory ECGs despite more false positives; leaves open outcome impact in prospective use.

Key Points

  • To evaluate the effectiveness of an AI model for direct-to-physician reporting of ambulatory ECGs compared to human technicians.
  • Tested an AI algorithm named DeepRhythmAI on 14,606 ECG recordings.
  • Compared AI performance with human technicians using a sample of 5,235 rhythm events.
  • Involved annotations by 17 cardiologist consensus panels for critical arrhythmia detection.
  • AI model sensitivity for critical arrhythmias was 98.6%, versus 80.3% for technicians.
  • AI model had a false-negative rate of 3.2/1,000 patients compared to 44.3/1,000 for technicians.
  • AI produced a higher false-positive rate of 12/1,000 patient days compared to 5/1,000 for technicians.

Structured PICO

Does an ensemble AI model (DeepRhythmAI) improve the identification of critical arrhythmias in ambulatory ECG recordings compared to certified ECG technicians?

P
Population
14,606 individual ambulatory ECG recordings (mean duration = 14 ± 10 days). A random sample of 5,235 rhythm events (2,236 critical arrhythmias) was selected for validation.
I
Intervention
Ensemble AI model (DeepRhythmAI) for beat-by-beat annotation of ambulatory ECGs
C
Comparator
Certified ECG technicians (n = 167)
O
Outcome
Sensitivity for the identification of critical arrhythmias (reference standard: 17 cardiologist consensus panels)surrogate

The DeepRhythmAI model demonstrates superior sensitivity and significantly reduces false-negative critical arrhythmia diagnoses compared to human technicians, though with a modest increase in false positives.

Cite This Study

Johnson et al. (2025) studied this question. DeepRhythmAI detected critical arrhythmias with 98.6% sensitivity versus 80.3% for technicians, reducing false-negatives by over 10-fold but increasing false-positives.

synapsesocial.com/papers/699a9e2d482488d673cd4aaahttps://doi.org/10.5167/uzh-292031
View Full Paper
Ask AI
Bookmark
Share